- Email: [email protected]
Cloud point–dispersive μ-solid phase extraction of hydrophobic organic compounds onto highly hydrophobic core–shell Fe2O3@C magnetic nanoparticles
Journal of Chromatography A, 1251 (2012) 33–39
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Cloud point–dispersive -solid phase extraction of hydrophobic organic compounds onto highly hydrophobic core–shell Fe2 O3 @C magnetic nanoparticles Dimosthenis L. Giokas a,∗ , Qing Zhu b , Qinmin Pan b,∗ , Alberto Chisvert c a b c
Laboratory of Analytical Chemistry, Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece School of Chemical Engineering and Technology and State Key Laboratory of Robotics and System, Harbin Institute of Technology, Harbin 150001, China Departamento de Química Analítica, Facultad de Química, Universidad de Valencia, 46100 Valencia, Spain
a r t i c l e
i n f o
Article history: Received 15 May 2012 Received in revised form 15 June 2012 Accepted 16 June 2012 Available online 26 June 2012 Keywords: Cloud point extraction Dispersive micro-solid phase extraction Highly hydrophobic magnetic nanoparticles Ultrasound-assisted back-extraction UV filters
a b s t r a c t A novel two-step extraction technique combining cloud point extraction (CPE) with dispersive microsolid phase extraction (D--SPE) is presented in this work for the first time. The method involves initial extraction of the target analytes by CPE in the micelles of a non-ionic surfactant medium; then highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C magnetic nanoparticles (MNPs) are used to retrieve the micellar phase. In that manner, the micellar phase containing the analytes is the target of the D--SPE step rather than the analytes directly. MNPs are then collected by the application of an adscititious magnetic field overcoming the need for specific steps associated with CPE such as centrifugation to separate the surfactant-rich phase, refrigeration of the condensed micellar phase to reduce its viscosity or appropriate apparatus that enable direct sampling of the surfactant-rich phase. A noteworthy feature of the method is the introduction of highly oleophilic MNPs, which afford rapid and quantitative mass transfer of the surfactant phase, as opposed to other more conventional hydrophobic nanoparticles. In that manner, fast and reproducible extraction is accomplished, lending improved analytical features compared to conventional CPE, such as reduced analysis time and relative inertness to surfactant concentration and equilibration temperature. The analytes were recovered from the surface of MNPs by ultrasound-assisted back-extraction in a water-immiscible organic solvent where analytes are readily partitioned but the surfactant has limited solubility, thus minimizing its interference during chromatographic detection. As an analytical demonstration, different UV absorbing chemicals with various physico-chemical properties were used as model organic compounds for optimizing the parameters associated with this novel two-step extraction approach. The proposed method, combining two different and efficient techniques, offers satisfactory analytical features in terms of repeatability (4.5–7.5%), reproducibility (7.0–14.9%) and accuracy (88.5–97.2%). Most importantly it poses as an alternative and fast method for sample pretreatment opening new insights in surfactant-mediated extractions. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Sample preparation is considered as the most critical step in overall analytical process since it has a multifarious role related to analyte extraction, preconcentration and clean-up from co-existing species. As a result, a large number of research efforts have been devoted to sample preparation and a wide array of analytical procedures have been developed, ranging in complexity from manual process to highly sophisticated mechanical devices. In the course of this endeavor, research on conventional methods like liquid–liquid extraction (LLE) and solid-phase extraction (SPE) has been replaced by procedures such as cloud point extraction (CPE) [1], dispersive solid phase extraction (DSPE) [2], stir bar sorptive extraction
∗ Corresponding authors. E-mail addresses: [email protected] (D.L. Giokas), [email protected] (Q. Pan). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.06.054
(SBSE) [3], membrane assisted solvent extraction (MASE) [4], as well as microextraction methods such as solid phase microextraction (SPME) [5], liquid phase microextraction (LPME) and related techniques [6] and recently dispersive micro-solid phase extraction (D--SPE) [7]. All these methods offer many advantages such as the use of small volumes of non-toxic organic solvents, high reproducibility, fast sample treatment and compatibility with most analytical instrumentation. However, although each method offers certain advantages it also encounters specific limitations, a feature that gives rise to the incessant efforts to develop new procedures or reinstate conventional methods resolving their drawbacks. In this direction, the combination of different sample preparation methods in the same experimental procedure has recently gained more attention. To date, research on the topic is still limited and concerns the combined use of SPE with dispersive liquid–liquid microextraction (DLLME) [8], DSPE combined with DLLME [9], DLLME followed by D--SPE [10], accelerated solvent
34
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
extraction (ASE) coupled to DSPE [11], matrix solid-phase dispersion (MSPD) combined with ultrasound-assisted DLLME [12]. From these studies, it can be inferred that combined extraction methods are very promising in minimizing or overcoming certain limitations of each individual technique, accomplishing improved selectivity, amending recoveries and ameliorating enhancement factors while alleviating time-consuming and laborious enrichment process of loading large sample volumes [7]. Among the most popular methods of the past decade, that could benefit from combined sample preparation procedures are those resorting to the use of supramolecular solvents (SUPRAS), namely surfactant-mediated extractions [13]. SUPRAS-based methods are highly versatile and can be tailored depending on the intended application or detection device. They afford high preconcentration factors, reduced toxicity, relatively fast analysis time and wide range of extraction solvents (i.e., surfactants) with different functionalities (ionic, hydrophobic or combined). However, despite their large versatility, they are subjected to certain limitations. In the majority of these methods centrifugation is required to separate the donor phase (i.e., sample) from the acceptor phase (i.e., surfactant) [1,13], which can be time consuming when dealing with large sample volumes. Another issue of concern is the collection of the surfactant-rich phase. Cooling is a common approach to increase its viscosity and facilitate its separation from the sample by decantation. On the other hand, reproducible retrieval of the surfactant phase by direct sampling requires appropriate apparatus such as conical or narrow-neck centrifuge vials to collect the surfactant phase. Last but not the least, the concentration of surfactant used for extraction commonly determines the analytical signal response since increasing surfactant concentrations in the extraction step induce a concomitant increase in the surfactant-rich phase volume remaining after extraction, which results in lower signals due to dilution [1,14]. All these procedures increase analysis time and experimental effort and are in certain occasions prone to the analysis’s efficiency. In this respect, and considering the widespread use of SUPRAS-mediated extractions, expansion of the overall procedure would be beneficial, especially if accompanied by simplification, reduced analysis time and relieve the technique, to the extent possible, from inherent limitations of the experimental procedure. Based on this rationale, the aim of this work is to reinstate micelle-mediated extraction resorting to the benefits afforded by magnetic nanoparticles. To this end, a two-step extraction technique relying on a new approach that involves the combined use of CPE and D--SPE is proposed. CPE of different UV absorbing chemicals, as model organic compounds, is initially applied as means to isolate them from the bulk sample phase. Then, the micellar phase, containing the analytes, is retrieved through D--SPE onto the surface of highly hydrophobic MNPs. The target analytes are back-extracted into a water immiscible organic solvent which is analyzed directly by liquid chromatography. The method overcomes the need for time consuming steps associated with CPE, such as centrifugation or cooling, does not require special apparatus and minimizes certain limitations encountered in typical CPE-procedures. To our knowledge, this is the first time that CPE is applied as a microscale sample preparation method, giving rise to the potential of reinstating SUPRAS-based extraction techniques.
2. Experimental 2.1. Chemicals and reagents Analytical-grade iron (II) sulfate heptahydrate (FeSO4 ·7H2 O), terephthalic acid, lithium hydroxide monohydrate (LiOH·H2 O) and vinyl triethoxysilane, used for the synthesis of highly hydrophobic
polysiloxane-coated core–shell Fe2 O3 @C MNPs were purchased from Sigma–Aldrich (Steinheim, Germany). For the preparation of cobalt ferrite coated with oleic acid nanoparticles (CoFe2 O4 @oleic acid), cobalt (II) chloride hexahydrate (CoCl2 ·6H2 O) (Scharlau) iron (III) chloride hexahydrate (FeCl3 ·6H2 O) (Sigma–Aldrich) and oleic acid (Sigma–Aldrich) were used. Sodium hydroxide and hydrochloric acid used for synthesis and for adjusting the pH of the solutions were purchased from Merck (Darmstadt, Germany). UV filter compounds (sunscreens) namely 2-phenylbenzimidazole-5-sulfonic acid (PBS), 2-hydroxy-4-methoxybenzophenone (benzophenone3, BZ3), 3-(4-methylbenzyldene)-camphor (MBC), ethylhexyl 4-methoxycinnamate (EMC) and butyl methoxy dibenzoylmethane (BDM) used as model organic compounds were purchased from Merck. Standard stock solutions of each model compound (500 mg L−1 ) were prepared weakly in methanol and stored in dark containers at 4 ◦ C. Multicomponent working standard solutions were freshly prepared daily by proper dilution of the methanolic stock standard solutions. The non-ionic surfactant Triton X-114 (TX-114) and the anionic surfactant sodium dodecyl sulfate (SDS) were obtained from Fluka (Buchs, Switzerland) and were used without further purification. HPLC-grade methanol, water, butanol, acetone, acetonitrile, propanol, ethanol, toluene, n-hexane, and dichloromethane were obtained from Fisher (Loughborough, UK). 2.2. Synthesis and characterization of magnetic nanoparticles Highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs varying from 30 to 200 nm in size coated with a carbon layer of 7–36 nm thickness were fabricated according to Zhu et al. [15]. Briefly, 8 mmol of terephthalic acid and 16 mmol of lithium hydroxide monohydrate were completely dissolved in 150 mL of distilled water. Then, 8 mmol of iron (II) sulfate heptahydrate salt was added under stirring at ambient temperature for 5 h to obtain the ferric benzoate precursor. The preparation of Fe2 O3 @C nanoparticles was conducted by sintering of the precursor in sealed quartz tubes at 600 ◦ C for 6 h. The synthesized Fe2 O3 @C nanoparticles were coated with low-surface-energy polysiloxane layers in an ethanol solution containing 2 wt% vinyl triethoxysilane for 3 h. The obtained powder was filtered and dried at 140 ◦ C for another 8 h. The synthetic procedure for the preparation of cobalt ferriteoleic acid coated (CoFe2 O4 @oleic acid) MNPs with an average diameter of 5–20 nm was replicated from a previous work [16]. More specifically, 125 mL of 0.4 M FeCl3 solution and 125 mL of 0.2 M CoCl2 solution were mixed under continuous stirring. Then 125 mL of 3 M sodium hydroxide solution were added dropwise. Next, 10 mL of oleic acid was added and the reaction mixture was heated to 80 ◦ C for 1 h. Finally, the MNPs were separated from the bulk solution by magnetic decantation, washed several times with ultrapure water and ethanol and air-dried. Meticulous characterization of the synthesized MNPs can be found in previous works [15,16], where the interested reader is referred for more detailed reading. 2.3. Extraction procedure CPE coupled to magnetic solid phase separation was performed by batch equilibrium technique using aqueous standard solutions (up to 50 mL) containing the target analytes. The solutions were initially spiked with Triton X-114 (10%, w/v) to yield a final concentration of 0.1% (w/v). The pH of the solution was adjusted to the value of 4 with dilute HCl, mixed in a vortex and kept for 5 min in a thermostated bath at 40 ◦ C with interim mixing. Then, 10 mg of polysiloxane-coated core–shell Fe2 O3 @C MNPs powder was added into the solution. The mixture was vortexed for 30 s and left to stand for 1 min. The MNPs were rapidly collected on the vessel wall under the influence of an adscititious magnetic field provided by a strong
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
35
Fig. 1. Simplified scheme of the two-step extraction process (CPE-D--SPE).
Nd–Fe–B magnet (0.55 T) and the aqueous phase was magnetically decanted. Excess water was removed by passing a nitrogen stream at ∼1 bar for 2 min. Desorption of the surfactant phase as well as the target analytes was accomplished by washing the sorbent twice with 250 L of dichloromethane and accelerated by ultrasonication for 30 s. The extract was recuperated in Eppendorf vials by magnetic decantation and directly injected into the LC-UV/diode-array detector system for analysis. A schematic of the two-step extraction process is shown in Fig. 1. 2.4. Real samples Genuine water samples were collected from aquatic ecosystems (river subject to human discharges and eutrophicated lake) and filtered to remove any suspended solids through a Whatman 0.45 m filter under vacuum. The samples were stored at 4 ◦ C until use for no more than a week. 2.5. Liquid chromatographic analysis The LC system comprised a Shimadzu LC-10AD high-pressure solvent delivery pump, with a 20-L Rheodyne sample loop injector and a Shimadzu SPD-M20A UV/diode-array detector. The analytical column was an Eclipse XDB-C18, with 5 m particles (15 cm × 4.6 mm I.D.). The column temperature was maintained at 30 ◦ C. Water (100 mM SDS)/acetonitrile (20/80%, v/v) was used for the isocratic elution of the analytes [17]. Peak areas were recorded at 305 nm except for BDM which was measured at 355 nm. A typical chromatogram is demonstrated in Fig. S1 (see Supporting Information).
(Fig. S2 and Video demonstration B – Supporting Information). These observations show that plain hydrophobic coatings did not enable the collection of the micellar aggregates onto the surface of MNPs while super-hydrophobic coatings afforded an excellent medium that favored this process. Based on these findings, the use of highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs was selected for the D--SPE step. Another significant parameter was the selection of the appropriate solvent that could overcome the strong adherence to the highly oleophilic surface and back-extract the analytes. In this context, preliminary tests were run by comparing water miscible solvents like methanol, ethanol, acetone, propanol butanol and acetonitrile with water immiscible solvents such as n-hexane, dichloromethane and toluene. Comparative projection of peak areas revealed that water immiscible solvents offered significantly superior extraction for the more hydrophobic compounds (BZ3, BDM, MBC and EMC) (Fig. 2) and lower extraction of the surfactant (not shown). However, PBS was not back-extracted, as no detectable peak was observed. On the contrary, water miscible solvents enabled back-extraction of PBS suffered from poor extraction of the more hydrophobic analytes and higher extraction of the surfactant. These results concur with previous studies, which exploited the low solubility of non-ionic surfactants in water immiscible solvents, to back-extract the analytes while minimizing surfactant interference in both liquid and gas chromatographic analysis [17,18]. On the other hand, the fact that PBS was not back-extracted is attributed to its highly hydrophilic nature, a property that does not favor its back-extraction into a hydrophobic organic phase. These observations indicate that the method offers good selectivity for the determination of hydrophobic organic compounds; therefore, PBS was omitted from the subsequent study. Taking into
3. Results and discussion 3.1. Preliminary experiments Before proceeding with method development, the optimization of basic parameters that play a crucial role to the overall performance and applicability of the method was investigated. For the convenience of the analysis and for minimizing reagents consumption, 5 mL aqueous standard solutions containing each analyte at 1 g/mL were used for the optimization study. The first factor was the selection of the appropriate nanoparticle sorbent. Two types of hydrophobic ferromagnetic nanoparticles were examined; the hydrophobic CoFe2 O4 @oleic acid and the highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs. The superiority of the latter was evident already from the first experiments. The first observation is that polysiloxanecoated core–shell Fe2 O3 @C MNPs spontaneously disperse into the cloudy surfactant solution (Video demonstration A – Supporting Information) suggesting that surfactant coating renders these MNPs hydrophilic, an attribute that facilitates their rapid and facile mixing into the aqueous phase, which is otherwise impossible [15]. Application of an external magnetic field revealed that they were able to completely remove the micellar phase, yielding a clear (transparent) solution, as opposed to CoFe2 O4 @oleic acid MNPs
Fig. 2. Influence of different desorption solvents on the back-extraction of the model analytes. MeOH: methanol; EtOH: ethanol; n-PAOH: n-propanol; MeCN: acetonitrile; Acet: acetone; n-BuOH: n-butanol; DCM: dichloromethane.
36
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Fig. 3. Influence of solution pH on the analytical signal response.
consideration the higher signal intensities observed with CH2 Cl2 it was selected as the optimum extraction solvent. 3.2. Optimization of CPE parameters Due to the non-polar nature of the extraction medium, neutral (i.e., not ionized) forms of the analytes should be preferably extracted. Based on the pKa values of the analytes BZ3 (pKa = 7.56) should preferably be extracted at acidic pH conditions while for the other analytes the influence of pH shall be less important since the partition coefficients of MBC and EMC do not change with pH, while the pKa value of BDM is close to 9.74. Although there is a plethora of literature reports on the optimum pH for the extraction of UV absorbing chemicals [16,17,20,21] the presence of micellar medium can alter the extent of an analyte’s ionization compared to that in its absence [19]. Therefore, the influence of pH on the extraction efficiency was studied in the pH range between 2 and 8. The results depicted in Fig. 3 show that acidic conditions are favorable for extraction, while at higher pH values hydrolysis conceivably contributed to the declining extraction yield. Therefore, acidic conditions (pH = 4) were applied. Since CPE is the first stage of the two-step extraction process it has a significant role in retrieving the analytes from the aqueous phase. Therefore, there must be an adequate excess of micelles to quantitatively extract the analytes. For the purposes of this study the effect of surfactant availability on the extraction performance was investigated in the range of 2.5–40 mg (or 0.05–0.8%, w/v). The graph depicted in Fig. 4 shows that the signal experiences an increase up to 5 mg (0.1%, v/v) TX-114. At higher amounts an interesting pattern is observed. The signal initially undergoes a slight decrease and then either reaches a plateau or exhibits a slightly increasing profile. To shed light on this phenomenon the concentration of Triton X-114 remaining after D--SPE was monitored by UV/Vis spectrophotometry at 275 nm. The amount of surfactant absorbed onto the surface of 10 mg nano-sorbent was increasing up to 5 mg. At higher concentrations, the amount of surfactant remaining in solution after D--SPE, was above its critical micellar concentration (c.m.c.) suggesting the presence of micelles in the bulk aqueous phase. A plausible assumption is that these micelles compete for the target analytes and this justifies the reduction in the extraction yield observed at TX-114 amount higher than 5 mg. However, placement of excess of surfactant does not induce further reduction suggesting that other mechanisms are also effective. According to previous studies, non-ionic surfactant absorption onto hydrophobic surfaces result in the formation of a monolayer,
Fig. 4. Effect of surfactant (TX-114) concentration on the extraction.
with the ethylene oxide groups in contact with the solution [22], which is attributed to hydrophobic interactions between the surfactants and nanoscale surface [23]. As the concentration increases non-ionic surfactants may start to form aggregates or micelles (i.e., admicelles) of hemicylinder structure on the surface [24,25], which increases the amount of surfactant partitioned on the surface and further increases the dispersion of nanoparticles. Therefore, the concentration of free micelles in solution is reduced favoring analytes extraction on the nanoparticle surface. It should be noted that under the experimental conditions applied (40 ◦ C and 0.1% NaCl) cloud point phase separation is favored which means that the repulsion between the hydrophilic groups of micelles is screened due to dehydration and salting out phenomena [1,13,14]. Therefore, it is possible that free micelles gradually interact with surfactant monomers or admicelles absorbed on the surface of MNPs following a mechanism that resembles the mechanism of aggregation responsible for the cloud point phenomenon. The fact that the amount of surfactant in the final extract increased with increasing initial concentration concurs with the above mechanisms. From an analytical point of view these results show that the procedure does not exhibit a strong dependence on surfactant concentration as opposed to classic CPE (where increase in surfactant concentration results in lower signals due to increased surfactant-rich phase volume). This attribute favors extraction yet high working surfactant concentrations result in increased surfactant residues in the final extract. Depending on the chromatographic elution program, increased surfactant concentrations may eventually interfere with the resolution of the target analytes although this was not a limitation in our work. All things considered, 5 mg of TX-114 (0.1%, w/v) was used throughout. The peak areas of the four model analytes were found to increase with increasing temperature up to 40 ◦ C with only insignificant gains attained up to 70 ◦ C (not shown). At higher temperatures (70–100 ◦ C) extraction yield gradually attenuated possible due to thermal instability of the surfactant aggregates or acceleration of the hydrolysis rates of the compounds [17]. The temperature for optimum extraction was lower than those commonly reported for CPE applications in aqueous samples [13] as well as below 60 ◦ C, observed in direct CPE of the same analytes [17] which is a result of the post-CPE/D--SPE step. In typical CPE increased temperatures are advantageous because they disrupt the surfactant hydrogen bonds, causing dehydration of the micellar aggregates’ inner core,
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Fig. 5. Effect of incubation time during the CPE step on signal intensities.
leading to increased preconcentration efficiency by minimizing the surfactant-rich phase volume [14]. However, due to the D-SPE step, the surfactant is retrieved onto the surface of MNPs and back-extracted to a minimal extend into the CH2 Cl2 phase; thus, the analytical performance of the method is absolved from its dependence on the surfactant-rich phase volume, as opposed to conventional CPE procedures. Ionic strength generally has a beneficial effect in CPE because it reduces the cloud point temperature and favors the partitioning of hydrophobic organic compounds into the micellar core by modifying the solvation environment [1]. However, ionic strength regulation in the range of 0.1–3.0% (w/v) with NaCl had no influence on the analytical response of all compounds possibly due to the highly hydrophobic interface formed between surfactant tail and nanoparticle surface that favors non-polar interactions. A minor improvement was observed for the least hydrophobic of the model compounds examined, i.e., BZ3, at NaCl concentrations as low as 0.1% with only insignificant gains be attained at higher concentrations. The application of 0.1% NaCl was finally decided. The importance of incubation (heating) time of the samples at 40 ◦ C was recorded in a time span from 0 to 20 min. The profile presented in Fig. 5 reveals that the signal intensities of the target compounds remain unchanged after 5 min which is lower than 10 min, reckoned to be the minimum universal optimum for CPE [1,19,26], and lower than the 15 min reported for the CPE of the specific compounds [17]. This observation may be ascribed to the rapid mass transfer of the surfactant on the highly hydrophobic surface of the MNPs, facilitating analytes partitioning on the surfactant–nanoparticle interface through potent hydrophobic interactions. On the contrary, in typical CPE procedures, adequate equilibration time is necessary to ensure efficient partitioning of analytes into the micellar core in order to afford quantitative extraction.
Fig. 6. Influence of variable amounts of polysiloxane-coated core–shell Fe2 O3 @C MNPs on analyte peak areas.
Increasing the extraction time for the D--SPE step (i.e., mixing time of MNPs with the sample) showed an increase in signal intensities up to 30 s reaching a plateau thereafter (Fig. 7). Interestingly, partial extraction was accomplished even when no mixing was applied which is attributed to the spontaneous dispersion of MNPs in the micellar solution, as demonstrated previously. An extraction time of 30 s was therefore assigned to the D--SPE step although higher extraction times can be safely deployed only at the expense of increasing analysis time. Agitation of the MNPs was performed at maximum speed to facilitate fast partitioning of the surfactant phase on the MNPs. Decreasing agitation speed from 2800 to 200 rpm had a negative effect on the extraction yield of the four compounds. Nevertheless, mixing time was a key factor determining the influence of agitation speed. In fact, decreasing vortex speed could be counterbalanced by an increase in mixing time; for the convenience of the analysis though, maximum mixing rate was applied. The desorption time necessary to back-extract the target analytes into the dichloromethane phase was subsequently tested (Fig. S3 – Supporting Information). Extraction was assisted by
3.3. Optimization of D--SPE conditions The first parameter investigated for its influence on the postCPE/D--SPE process was the amount of MNPs necessary to afford quantitative extraction. As shown in Fig. 6, increasing the amount of MNPs from 10 to 60 mg (2–12 mg/mL), the extraction efficiency was enhanced for the more hydrophobic EMC (log Kow = 5.80) and MBC (log Kow = 5.47) while the opposite trend was found for the less hydrophobic BZ3 (log Kow = 3.79) and BDM (log Kow = 4.51) [27]. Based on this pattern, 10 mg of the magnetic nano-sorbent was employed in the following experiments.
37
Fig. 7. Analyte signal response to extraction (mixing) time.
38
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Table 1 Recoveries and relative standard deviation (n = 3) of the model organic compounds from spiked water samples. 50 ng/mL
100 ng/mL
River water
BZ-3 MBC BDM EMC
Lake water
River water
Lake water
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
97.2 93.0 88.5 94.0
5.2 7.2 7.1 5.2
94.3 93.5 85.5 92.5
5.8 6.5 6.7 5.1
95.9 96.1 90.2 95.4
4.5 6.8 6.5 4.6
92.2 90.8 87.8 91.3
5.0 6.0 6.1 4.6
ultrasonic irradiation in order to accelerate analyte transfer [17,18]. The desorption/ultrasonication time was not found to have any influence on the analytical signal response above 30s. To ensure fast analysis time a desorption time of 30 s was therefore adopted. The desorption of the target analytes from the MNPs surface was investigated by varying elution solvent volume from 250 to 2000 L and for 1–5 runs. Increasing the elution volume was found to induce a decline in the analytical signals proportional to the theoretical preconcentration factor suggesting the lack of influence on the extraction performance. The use of lower solvent volume was not feasible because ultrasonication resulted in random scattering of the MNPs in the vial walls. Therefore, 250 L was decided as optimum. The peak intensities observed from 5 consecutive extractions revealed that two consecutive extractions are necessary to back-extract the analytes since approximately 10% of the analytes were detected in the second extraction step. Additional steps did not offer any improvement. The maximal sample volume that can be extracted with the proposed procedure was investigated by analyzing 10–100 mL of aqueous standard solutions considering that insufficient extraction yield occurs when it is decreases below 90%. This condition was satisfied up to 50 mL which is the maximum sample volume that can be extracted yielding theoretical preconcentration factors of approximately 100 (or higher with solvent evaporation). Extraction of larger sample volumes is feasible by appropriately scaling-up the experimental variables. This was evidenced by analyzing 100 mL sample volume with increased agitation time (2 min) and MNPs amount (30 mg). To ensure that the volume of CH2 Cl2 will not be a limiting factor, due to the increased amount of MNPs, it was adjusted to 1 mL (2× 500 L). Satisfactory extraction yields (88.7–93.1%) were observed suggesting that the method is flexible and can be tailored to cope with the requirements of the analysis. However, for the analytical demonstration purposes of this study, no further efforts were devoted to optimize the scaling-up factors of the experimental variables for the analysis larger sample volumes. The re-usability of the sorbent was finally assessed in a series of aqueous standard solutions (n = 5) extracted under the aforementioned procedure. After extraction, the MNPs were washed twice with ethanol and CH2 Cl2 under ultrasonic irradiation for 1 min. The solvent was removed by magnetic decantation and MNPs were dried in a ventilated oven at 50 ◦ C. The sorbents could be re-used for up to 4 times without impairing the extraction efficiency. It should be pointed out that MNPs re-usability strongly depended on the homogenization of the sorbent. Good homogenization was necessary to avoid agglomerates formed after drying. Finally, it should be emphasized that quantitative recovery of the sorbent was not feasible since a part strongly adhered to the vials after drying. Therefore, re-use after the fourth cycle offers insignificant gains with regard to sorbent consumption. On the basis of the above optimization study the total analysis time under the optimal extraction conditions for the two-step CPE-D--SPE process was less than 9 min which is a significant advantage over conventional CPE procedures where at least
30 min are required, depending also on the capacity of the centrifuge. 4. Analytical features and method application The analytical figures of merit of the proposed method were determined by using 10 mL aqueous standard solutions of the target analytes over a concentration range of 10–1000 ng/mL for BZ3, MBC and EMC and 50–1000 ng/mL for BDM. Method calibration was performed by plotting the mean peak area obtained from three measurements versus sample concentration. The regression coefficients varied between 0.990–0.997 and the detection limits, calculated as three times the signal-to-noise ratio were 1.43 ng/mL for BZ3, 2.4 ng/mL for MBC, 7.5 ng/mL for BDM and 1.94 ng/mL for EMC. Further improvement can be attained by resorting to the use of more sensitive analytical techniques, like LC–MS or GC–MS, by extracting larger sample volumes or by evaporating the extraction solvent to lower volume. However, both evaporation under N2 and at 50 ◦ C in a ventilated oven, were proven inefficient since an average loss of 30% in signal peak areas of BZ3, MBC and EMC, compared to direct analysis, was recorded; in addition, BDM peak disappeared. Mild evaporation of the elution solvent to dryness in a centrifugal concentrator under vacuum at room temperature was found to be the most appropriate technique, minimizing analytes’ loss and coping to a large extent with BDM evaporation. A series of experiments were then performed in order to assess the precision, repeatability and accuracy of the method. The repeatability of the method, expressed as intra-day relative standard deviation (RSD) of eight replicates was 4.5% for BZ3, 6.2% for MBC and 5.2% for EMC at 20 ng/mL and 7.5% and BDM at 50 ng/mL concentration level. The reproducibility of the measurements, expressed as inter-day RSD of five replicates was 8.7% for BZ3, 11.5% for MBC and 7.0% for EMC at 20 ng/mL and 14.9% for BDM at 50 ng/mL, which was deemed as satisfactory. Finally, the accuracy of the method was evaluated by spiking two genuine water samples at two different concentration levels (50 ng/mL and 100 ng/mL). The observed recoveries (Table 1) ranged between 85.5 and 97.2% suggesting that the method is tolerant of matrices generally representative of environmental water samples. 5. Conclusions In this study, a new two-step extraction technique combining CPE and D--SPE mediated by highly hydrophobic polysiloxanecoated core–shell Fe2 O3 @C magnetic nanoparticles is presented. On the basis of the results obtained the method affords significant simplification to conventional CPE procedures, without resource to appropriate or specialized apparatus, alleviating the need for specific sample handling treatment such as centrifugation or freezing of the surfactant-rich phase, significantly abridges the overall analysis time (at least 3-fold) and minimizes the importance of CPErelated parameters (i.e., surfactant concentration and incubation
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
temperature) on the extraction. As an analytical demonstration, the parameters associated with method performance were optimized using four UV absorbing chemicals as model organic compounds and applied to spiked environmental water samples yielding good analytical features in terms of accuracy (85.5–97.2%), repeatability (4.5–7.5%) and reproducibility (7.0–14.9%). Considering the wide array of supramolecular solvents, offering a variety of functionalities and the high versatility of magnetic nanoparticles coatings, this study opens new venues in reinstating surfactant-mediated extractions. The use of ionic surfactant-mediated extractions that completely alleviate surfactant interference in chromatographic analysis, extraction of hydrophobic metal chelates and method automation are prosperous fields of research. Work along this line is in progress. Acknowledgements This research has been co-financed by the European Union (European Social Fund–ESF) and Greek national funds through the Operational Program “Education and Lifelong Learning” of the National Strategic Reference Framework (NSRF)–Research Funding Program: THALIS. Investing in knowledge society through the European Social Fund. The preparation of highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C magnetic nanoparticles was financially supported by the Natural Science Foundation of China (NSFC, Grant No. 50803013). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2012.06.054.
39
References [1] E.K. Paleologos, D.L. Giokas, M.I. Karayannis, Trends Anal. Chem. 24 (2005) 426. [2] L. Ramos, J. Chromatogr. A 1221 (2011) 84. [3] F. Sánchez-Rojas, C. Bosch-Ojeda, J.M. Cano-Pavón, Chromatographia 69 (2009) 79. [4] B. Hauser, M. Schellin, P. Popp, Anal. Chem. 76 (2004) 6029. [5] Spietelun, M. Pilarczyk, A. Kloskowski, J. Namie´snik, Chem. Soc. Rev. 39 (2010) 4524. [6] D. Han, K.H. Row, Microchim. Acta 176 (2012) 1. [7] M. Cruz-Vera, R. Lucena, S. Cardenas, M. Valcarcel, Anal. Methods 3 (2011) 1719. [8] N. Fattahi, S. Samadi, Y. Assadi, M.R.M. Hosseini, J. Chromatogr. A 1169 (2007) 63. [9] Q. Wua, C. Wang, Z. Liu, C. Wu, X. Zeng, J. Wen, Z. Wang, J. Chromatogr. A 1216 (2009) 5504. [10] Z-G. Shi, H.K. Lee, Anal. Chem. 82 (2010) 1540. [11] R.Z. Yang, J.H. Wang, M.L. Wang, R. Zhang, X.Y. Lu, W.H. Liu, J. Chromatogr. Sci. 49 (2011) 702. [12] H. Wang, H. Yan, J. Qiao, J. Sep. Sci. 35 (2012) 292. [13] A.B. Gómez, M.D. Sicilia, S. Rubio, Anal. Chim. Acta 677 (2010) 108. [14] R. Carabias-Martinez, E. Rodriguez-Gonzalo, B. Moreno-Cordero, J.L. PerezPavon, C. Garcia-Pinto, E. Fernandez Laespada, J. Chromatogr. A 902 (2000) 251. [15] Q. Zhu, F. Tao, Q.M. Pan, ACS Appl. Mater. Interfaces 2 (2010) 3141. [16] I.P. Roman, A. Chisvert, A. Canals, J. Chromatogr. A 1218 (2011) 2467. [17] D.L. Giokas, V.A. Sakkas, T.A. Albanis, D.A. Lampropoulou, J. Chromatogr. A 1077 (2005) 19. [18] T.I. Sikalos, E.K. Paleologos, Anal. Chem. 77 (2005) 2544. [19] R.P. Frankewlcht, W.L. Hinze, Anal. Chem. 66 (1994) 944. [20] D.L. Giokas, V.A. Sakkas, T.A. Albanis, J. Chromatogr. A 1026 (2004) 289. [21] L. Vidal, A. Chisvert, A. Canals, A. Salvador, Talanta 81 (2010) 549. [22] L.M. Grant, F. Tiberg, W.A. Ducker, J. Phys. Chem. B 102 (1998) 4288. [23] Y. Bai, D. Lin, F. Wu, Z. Wang, B. Xing, Chemosphere 79 (2010) 362. [24] Y. Geng, M.Y. Liu, J. Li, X.M. Shi, J.K. Kim, Comp. A: Appl. Sci. Manuf. 39 (2008) 1876. [25] M. Soria-Sanchez, A. Maroto-Valiente, A. Guerrero-Ruiz, D.M. Nevskaia, J. Colloid Interface Sci. 343 (2010) 194. [26] C.D. Stalikas, Trends Anal. Chem. 21 (2002) 343. [27] US EPA, Estimation Programs Interface SuiteTM for Microsoft® Windows, v 4.10, United States Environmental Protection Agency, Washington, DC, USA, 2011.
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Cloud point–dispersive -solid phase extraction of hydrophobic organic compounds onto highly hydrophobic core–shell Fe2 O3 @C magnetic nanoparticles Dimosthenis L. Giokas a,∗ , Qing Zhu b , Qinmin Pan b,∗ , Alberto Chisvert c a b c
Laboratory of Analytical Chemistry, Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece School of Chemical Engineering and Technology and State Key Laboratory of Robotics and System, Harbin Institute of Technology, Harbin 150001, China Departamento de Química Analítica, Facultad de Química, Universidad de Valencia, 46100 Valencia, Spain
a r t i c l e
i n f o
Article history: Received 15 May 2012 Received in revised form 15 June 2012 Accepted 16 June 2012 Available online 26 June 2012 Keywords: Cloud point extraction Dispersive micro-solid phase extraction Highly hydrophobic magnetic nanoparticles Ultrasound-assisted back-extraction UV filters
a b s t r a c t A novel two-step extraction technique combining cloud point extraction (CPE) with dispersive microsolid phase extraction (D--SPE) is presented in this work for the first time. The method involves initial extraction of the target analytes by CPE in the micelles of a non-ionic surfactant medium; then highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C magnetic nanoparticles (MNPs) are used to retrieve the micellar phase. In that manner, the micellar phase containing the analytes is the target of the D--SPE step rather than the analytes directly. MNPs are then collected by the application of an adscititious magnetic field overcoming the need for specific steps associated with CPE such as centrifugation to separate the surfactant-rich phase, refrigeration of the condensed micellar phase to reduce its viscosity or appropriate apparatus that enable direct sampling of the surfactant-rich phase. A noteworthy feature of the method is the introduction of highly oleophilic MNPs, which afford rapid and quantitative mass transfer of the surfactant phase, as opposed to other more conventional hydrophobic nanoparticles. In that manner, fast and reproducible extraction is accomplished, lending improved analytical features compared to conventional CPE, such as reduced analysis time and relative inertness to surfactant concentration and equilibration temperature. The analytes were recovered from the surface of MNPs by ultrasound-assisted back-extraction in a water-immiscible organic solvent where analytes are readily partitioned but the surfactant has limited solubility, thus minimizing its interference during chromatographic detection. As an analytical demonstration, different UV absorbing chemicals with various physico-chemical properties were used as model organic compounds for optimizing the parameters associated with this novel two-step extraction approach. The proposed method, combining two different and efficient techniques, offers satisfactory analytical features in terms of repeatability (4.5–7.5%), reproducibility (7.0–14.9%) and accuracy (88.5–97.2%). Most importantly it poses as an alternative and fast method for sample pretreatment opening new insights in surfactant-mediated extractions. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Sample preparation is considered as the most critical step in overall analytical process since it has a multifarious role related to analyte extraction, preconcentration and clean-up from co-existing species. As a result, a large number of research efforts have been devoted to sample preparation and a wide array of analytical procedures have been developed, ranging in complexity from manual process to highly sophisticated mechanical devices. In the course of this endeavor, research on conventional methods like liquid–liquid extraction (LLE) and solid-phase extraction (SPE) has been replaced by procedures such as cloud point extraction (CPE) [1], dispersive solid phase extraction (DSPE) [2], stir bar sorptive extraction
∗ Corresponding authors. E-mail addresses: [email protected] (D.L. Giokas), [email protected] (Q. Pan). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.06.054
(SBSE) [3], membrane assisted solvent extraction (MASE) [4], as well as microextraction methods such as solid phase microextraction (SPME) [5], liquid phase microextraction (LPME) and related techniques [6] and recently dispersive micro-solid phase extraction (D--SPE) [7]. All these methods offer many advantages such as the use of small volumes of non-toxic organic solvents, high reproducibility, fast sample treatment and compatibility with most analytical instrumentation. However, although each method offers certain advantages it also encounters specific limitations, a feature that gives rise to the incessant efforts to develop new procedures or reinstate conventional methods resolving their drawbacks. In this direction, the combination of different sample preparation methods in the same experimental procedure has recently gained more attention. To date, research on the topic is still limited and concerns the combined use of SPE with dispersive liquid–liquid microextraction (DLLME) [8], DSPE combined with DLLME [9], DLLME followed by D--SPE [10], accelerated solvent
34
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
extraction (ASE) coupled to DSPE [11], matrix solid-phase dispersion (MSPD) combined with ultrasound-assisted DLLME [12]. From these studies, it can be inferred that combined extraction methods are very promising in minimizing or overcoming certain limitations of each individual technique, accomplishing improved selectivity, amending recoveries and ameliorating enhancement factors while alleviating time-consuming and laborious enrichment process of loading large sample volumes [7]. Among the most popular methods of the past decade, that could benefit from combined sample preparation procedures are those resorting to the use of supramolecular solvents (SUPRAS), namely surfactant-mediated extractions [13]. SUPRAS-based methods are highly versatile and can be tailored depending on the intended application or detection device. They afford high preconcentration factors, reduced toxicity, relatively fast analysis time and wide range of extraction solvents (i.e., surfactants) with different functionalities (ionic, hydrophobic or combined). However, despite their large versatility, they are subjected to certain limitations. In the majority of these methods centrifugation is required to separate the donor phase (i.e., sample) from the acceptor phase (i.e., surfactant) [1,13], which can be time consuming when dealing with large sample volumes. Another issue of concern is the collection of the surfactant-rich phase. Cooling is a common approach to increase its viscosity and facilitate its separation from the sample by decantation. On the other hand, reproducible retrieval of the surfactant phase by direct sampling requires appropriate apparatus such as conical or narrow-neck centrifuge vials to collect the surfactant phase. Last but not the least, the concentration of surfactant used for extraction commonly determines the analytical signal response since increasing surfactant concentrations in the extraction step induce a concomitant increase in the surfactant-rich phase volume remaining after extraction, which results in lower signals due to dilution [1,14]. All these procedures increase analysis time and experimental effort and are in certain occasions prone to the analysis’s efficiency. In this respect, and considering the widespread use of SUPRAS-mediated extractions, expansion of the overall procedure would be beneficial, especially if accompanied by simplification, reduced analysis time and relieve the technique, to the extent possible, from inherent limitations of the experimental procedure. Based on this rationale, the aim of this work is to reinstate micelle-mediated extraction resorting to the benefits afforded by magnetic nanoparticles. To this end, a two-step extraction technique relying on a new approach that involves the combined use of CPE and D--SPE is proposed. CPE of different UV absorbing chemicals, as model organic compounds, is initially applied as means to isolate them from the bulk sample phase. Then, the micellar phase, containing the analytes, is retrieved through D--SPE onto the surface of highly hydrophobic MNPs. The target analytes are back-extracted into a water immiscible organic solvent which is analyzed directly by liquid chromatography. The method overcomes the need for time consuming steps associated with CPE, such as centrifugation or cooling, does not require special apparatus and minimizes certain limitations encountered in typical CPE-procedures. To our knowledge, this is the first time that CPE is applied as a microscale sample preparation method, giving rise to the potential of reinstating SUPRAS-based extraction techniques.
2. Experimental 2.1. Chemicals and reagents Analytical-grade iron (II) sulfate heptahydrate (FeSO4 ·7H2 O), terephthalic acid, lithium hydroxide monohydrate (LiOH·H2 O) and vinyl triethoxysilane, used for the synthesis of highly hydrophobic
polysiloxane-coated core–shell Fe2 O3 @C MNPs were purchased from Sigma–Aldrich (Steinheim, Germany). For the preparation of cobalt ferrite coated with oleic acid nanoparticles (CoFe2 O4 @oleic acid), cobalt (II) chloride hexahydrate (CoCl2 ·6H2 O) (Scharlau) iron (III) chloride hexahydrate (FeCl3 ·6H2 O) (Sigma–Aldrich) and oleic acid (Sigma–Aldrich) were used. Sodium hydroxide and hydrochloric acid used for synthesis and for adjusting the pH of the solutions were purchased from Merck (Darmstadt, Germany). UV filter compounds (sunscreens) namely 2-phenylbenzimidazole-5-sulfonic acid (PBS), 2-hydroxy-4-methoxybenzophenone (benzophenone3, BZ3), 3-(4-methylbenzyldene)-camphor (MBC), ethylhexyl 4-methoxycinnamate (EMC) and butyl methoxy dibenzoylmethane (BDM) used as model organic compounds were purchased from Merck. Standard stock solutions of each model compound (500 mg L−1 ) were prepared weakly in methanol and stored in dark containers at 4 ◦ C. Multicomponent working standard solutions were freshly prepared daily by proper dilution of the methanolic stock standard solutions. The non-ionic surfactant Triton X-114 (TX-114) and the anionic surfactant sodium dodecyl sulfate (SDS) were obtained from Fluka (Buchs, Switzerland) and were used without further purification. HPLC-grade methanol, water, butanol, acetone, acetonitrile, propanol, ethanol, toluene, n-hexane, and dichloromethane were obtained from Fisher (Loughborough, UK). 2.2. Synthesis and characterization of magnetic nanoparticles Highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs varying from 30 to 200 nm in size coated with a carbon layer of 7–36 nm thickness were fabricated according to Zhu et al. [15]. Briefly, 8 mmol of terephthalic acid and 16 mmol of lithium hydroxide monohydrate were completely dissolved in 150 mL of distilled water. Then, 8 mmol of iron (II) sulfate heptahydrate salt was added under stirring at ambient temperature for 5 h to obtain the ferric benzoate precursor. The preparation of Fe2 O3 @C nanoparticles was conducted by sintering of the precursor in sealed quartz tubes at 600 ◦ C for 6 h. The synthesized Fe2 O3 @C nanoparticles were coated with low-surface-energy polysiloxane layers in an ethanol solution containing 2 wt% vinyl triethoxysilane for 3 h. The obtained powder was filtered and dried at 140 ◦ C for another 8 h. The synthetic procedure for the preparation of cobalt ferriteoleic acid coated (CoFe2 O4 @oleic acid) MNPs with an average diameter of 5–20 nm was replicated from a previous work [16]. More specifically, 125 mL of 0.4 M FeCl3 solution and 125 mL of 0.2 M CoCl2 solution were mixed under continuous stirring. Then 125 mL of 3 M sodium hydroxide solution were added dropwise. Next, 10 mL of oleic acid was added and the reaction mixture was heated to 80 ◦ C for 1 h. Finally, the MNPs were separated from the bulk solution by magnetic decantation, washed several times with ultrapure water and ethanol and air-dried. Meticulous characterization of the synthesized MNPs can be found in previous works [15,16], where the interested reader is referred for more detailed reading. 2.3. Extraction procedure CPE coupled to magnetic solid phase separation was performed by batch equilibrium technique using aqueous standard solutions (up to 50 mL) containing the target analytes. The solutions were initially spiked with Triton X-114 (10%, w/v) to yield a final concentration of 0.1% (w/v). The pH of the solution was adjusted to the value of 4 with dilute HCl, mixed in a vortex and kept for 5 min in a thermostated bath at 40 ◦ C with interim mixing. Then, 10 mg of polysiloxane-coated core–shell Fe2 O3 @C MNPs powder was added into the solution. The mixture was vortexed for 30 s and left to stand for 1 min. The MNPs were rapidly collected on the vessel wall under the influence of an adscititious magnetic field provided by a strong
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
35
Fig. 1. Simplified scheme of the two-step extraction process (CPE-D--SPE).
Nd–Fe–B magnet (0.55 T) and the aqueous phase was magnetically decanted. Excess water was removed by passing a nitrogen stream at ∼1 bar for 2 min. Desorption of the surfactant phase as well as the target analytes was accomplished by washing the sorbent twice with 250 L of dichloromethane and accelerated by ultrasonication for 30 s. The extract was recuperated in Eppendorf vials by magnetic decantation and directly injected into the LC-UV/diode-array detector system for analysis. A schematic of the two-step extraction process is shown in Fig. 1. 2.4. Real samples Genuine water samples were collected from aquatic ecosystems (river subject to human discharges and eutrophicated lake) and filtered to remove any suspended solids through a Whatman 0.45 m filter under vacuum. The samples were stored at 4 ◦ C until use for no more than a week. 2.5. Liquid chromatographic analysis The LC system comprised a Shimadzu LC-10AD high-pressure solvent delivery pump, with a 20-L Rheodyne sample loop injector and a Shimadzu SPD-M20A UV/diode-array detector. The analytical column was an Eclipse XDB-C18, with 5 m particles (15 cm × 4.6 mm I.D.). The column temperature was maintained at 30 ◦ C. Water (100 mM SDS)/acetonitrile (20/80%, v/v) was used for the isocratic elution of the analytes [17]. Peak areas were recorded at 305 nm except for BDM which was measured at 355 nm. A typical chromatogram is demonstrated in Fig. S1 (see Supporting Information).
(Fig. S2 and Video demonstration B – Supporting Information). These observations show that plain hydrophobic coatings did not enable the collection of the micellar aggregates onto the surface of MNPs while super-hydrophobic coatings afforded an excellent medium that favored this process. Based on these findings, the use of highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs was selected for the D--SPE step. Another significant parameter was the selection of the appropriate solvent that could overcome the strong adherence to the highly oleophilic surface and back-extract the analytes. In this context, preliminary tests were run by comparing water miscible solvents like methanol, ethanol, acetone, propanol butanol and acetonitrile with water immiscible solvents such as n-hexane, dichloromethane and toluene. Comparative projection of peak areas revealed that water immiscible solvents offered significantly superior extraction for the more hydrophobic compounds (BZ3, BDM, MBC and EMC) (Fig. 2) and lower extraction of the surfactant (not shown). However, PBS was not back-extracted, as no detectable peak was observed. On the contrary, water miscible solvents enabled back-extraction of PBS suffered from poor extraction of the more hydrophobic analytes and higher extraction of the surfactant. These results concur with previous studies, which exploited the low solubility of non-ionic surfactants in water immiscible solvents, to back-extract the analytes while minimizing surfactant interference in both liquid and gas chromatographic analysis [17,18]. On the other hand, the fact that PBS was not back-extracted is attributed to its highly hydrophilic nature, a property that does not favor its back-extraction into a hydrophobic organic phase. These observations indicate that the method offers good selectivity for the determination of hydrophobic organic compounds; therefore, PBS was omitted from the subsequent study. Taking into
3. Results and discussion 3.1. Preliminary experiments Before proceeding with method development, the optimization of basic parameters that play a crucial role to the overall performance and applicability of the method was investigated. For the convenience of the analysis and for minimizing reagents consumption, 5 mL aqueous standard solutions containing each analyte at 1 g/mL were used for the optimization study. The first factor was the selection of the appropriate nanoparticle sorbent. Two types of hydrophobic ferromagnetic nanoparticles were examined; the hydrophobic CoFe2 O4 @oleic acid and the highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C MNPs. The superiority of the latter was evident already from the first experiments. The first observation is that polysiloxanecoated core–shell Fe2 O3 @C MNPs spontaneously disperse into the cloudy surfactant solution (Video demonstration A – Supporting Information) suggesting that surfactant coating renders these MNPs hydrophilic, an attribute that facilitates their rapid and facile mixing into the aqueous phase, which is otherwise impossible [15]. Application of an external magnetic field revealed that they were able to completely remove the micellar phase, yielding a clear (transparent) solution, as opposed to CoFe2 O4 @oleic acid MNPs
Fig. 2. Influence of different desorption solvents on the back-extraction of the model analytes. MeOH: methanol; EtOH: ethanol; n-PAOH: n-propanol; MeCN: acetonitrile; Acet: acetone; n-BuOH: n-butanol; DCM: dichloromethane.
36
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Fig. 3. Influence of solution pH on the analytical signal response.
consideration the higher signal intensities observed with CH2 Cl2 it was selected as the optimum extraction solvent. 3.2. Optimization of CPE parameters Due to the non-polar nature of the extraction medium, neutral (i.e., not ionized) forms of the analytes should be preferably extracted. Based on the pKa values of the analytes BZ3 (pKa = 7.56) should preferably be extracted at acidic pH conditions while for the other analytes the influence of pH shall be less important since the partition coefficients of MBC and EMC do not change with pH, while the pKa value of BDM is close to 9.74. Although there is a plethora of literature reports on the optimum pH for the extraction of UV absorbing chemicals [16,17,20,21] the presence of micellar medium can alter the extent of an analyte’s ionization compared to that in its absence [19]. Therefore, the influence of pH on the extraction efficiency was studied in the pH range between 2 and 8. The results depicted in Fig. 3 show that acidic conditions are favorable for extraction, while at higher pH values hydrolysis conceivably contributed to the declining extraction yield. Therefore, acidic conditions (pH = 4) were applied. Since CPE is the first stage of the two-step extraction process it has a significant role in retrieving the analytes from the aqueous phase. Therefore, there must be an adequate excess of micelles to quantitatively extract the analytes. For the purposes of this study the effect of surfactant availability on the extraction performance was investigated in the range of 2.5–40 mg (or 0.05–0.8%, w/v). The graph depicted in Fig. 4 shows that the signal experiences an increase up to 5 mg (0.1%, v/v) TX-114. At higher amounts an interesting pattern is observed. The signal initially undergoes a slight decrease and then either reaches a plateau or exhibits a slightly increasing profile. To shed light on this phenomenon the concentration of Triton X-114 remaining after D--SPE was monitored by UV/Vis spectrophotometry at 275 nm. The amount of surfactant absorbed onto the surface of 10 mg nano-sorbent was increasing up to 5 mg. At higher concentrations, the amount of surfactant remaining in solution after D--SPE, was above its critical micellar concentration (c.m.c.) suggesting the presence of micelles in the bulk aqueous phase. A plausible assumption is that these micelles compete for the target analytes and this justifies the reduction in the extraction yield observed at TX-114 amount higher than 5 mg. However, placement of excess of surfactant does not induce further reduction suggesting that other mechanisms are also effective. According to previous studies, non-ionic surfactant absorption onto hydrophobic surfaces result in the formation of a monolayer,
Fig. 4. Effect of surfactant (TX-114) concentration on the extraction.
with the ethylene oxide groups in contact with the solution [22], which is attributed to hydrophobic interactions between the surfactants and nanoscale surface [23]. As the concentration increases non-ionic surfactants may start to form aggregates or micelles (i.e., admicelles) of hemicylinder structure on the surface [24,25], which increases the amount of surfactant partitioned on the surface and further increases the dispersion of nanoparticles. Therefore, the concentration of free micelles in solution is reduced favoring analytes extraction on the nanoparticle surface. It should be noted that under the experimental conditions applied (40 ◦ C and 0.1% NaCl) cloud point phase separation is favored which means that the repulsion between the hydrophilic groups of micelles is screened due to dehydration and salting out phenomena [1,13,14]. Therefore, it is possible that free micelles gradually interact with surfactant monomers or admicelles absorbed on the surface of MNPs following a mechanism that resembles the mechanism of aggregation responsible for the cloud point phenomenon. The fact that the amount of surfactant in the final extract increased with increasing initial concentration concurs with the above mechanisms. From an analytical point of view these results show that the procedure does not exhibit a strong dependence on surfactant concentration as opposed to classic CPE (where increase in surfactant concentration results in lower signals due to increased surfactant-rich phase volume). This attribute favors extraction yet high working surfactant concentrations result in increased surfactant residues in the final extract. Depending on the chromatographic elution program, increased surfactant concentrations may eventually interfere with the resolution of the target analytes although this was not a limitation in our work. All things considered, 5 mg of TX-114 (0.1%, w/v) was used throughout. The peak areas of the four model analytes were found to increase with increasing temperature up to 40 ◦ C with only insignificant gains attained up to 70 ◦ C (not shown). At higher temperatures (70–100 ◦ C) extraction yield gradually attenuated possible due to thermal instability of the surfactant aggregates or acceleration of the hydrolysis rates of the compounds [17]. The temperature for optimum extraction was lower than those commonly reported for CPE applications in aqueous samples [13] as well as below 60 ◦ C, observed in direct CPE of the same analytes [17] which is a result of the post-CPE/D--SPE step. In typical CPE increased temperatures are advantageous because they disrupt the surfactant hydrogen bonds, causing dehydration of the micellar aggregates’ inner core,
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Fig. 5. Effect of incubation time during the CPE step on signal intensities.
leading to increased preconcentration efficiency by minimizing the surfactant-rich phase volume [14]. However, due to the D-SPE step, the surfactant is retrieved onto the surface of MNPs and back-extracted to a minimal extend into the CH2 Cl2 phase; thus, the analytical performance of the method is absolved from its dependence on the surfactant-rich phase volume, as opposed to conventional CPE procedures. Ionic strength generally has a beneficial effect in CPE because it reduces the cloud point temperature and favors the partitioning of hydrophobic organic compounds into the micellar core by modifying the solvation environment [1]. However, ionic strength regulation in the range of 0.1–3.0% (w/v) with NaCl had no influence on the analytical response of all compounds possibly due to the highly hydrophobic interface formed between surfactant tail and nanoparticle surface that favors non-polar interactions. A minor improvement was observed for the least hydrophobic of the model compounds examined, i.e., BZ3, at NaCl concentrations as low as 0.1% with only insignificant gains be attained at higher concentrations. The application of 0.1% NaCl was finally decided. The importance of incubation (heating) time of the samples at 40 ◦ C was recorded in a time span from 0 to 20 min. The profile presented in Fig. 5 reveals that the signal intensities of the target compounds remain unchanged after 5 min which is lower than 10 min, reckoned to be the minimum universal optimum for CPE [1,19,26], and lower than the 15 min reported for the CPE of the specific compounds [17]. This observation may be ascribed to the rapid mass transfer of the surfactant on the highly hydrophobic surface of the MNPs, facilitating analytes partitioning on the surfactant–nanoparticle interface through potent hydrophobic interactions. On the contrary, in typical CPE procedures, adequate equilibration time is necessary to ensure efficient partitioning of analytes into the micellar core in order to afford quantitative extraction.
Fig. 6. Influence of variable amounts of polysiloxane-coated core–shell Fe2 O3 @C MNPs on analyte peak areas.
Increasing the extraction time for the D--SPE step (i.e., mixing time of MNPs with the sample) showed an increase in signal intensities up to 30 s reaching a plateau thereafter (Fig. 7). Interestingly, partial extraction was accomplished even when no mixing was applied which is attributed to the spontaneous dispersion of MNPs in the micellar solution, as demonstrated previously. An extraction time of 30 s was therefore assigned to the D--SPE step although higher extraction times can be safely deployed only at the expense of increasing analysis time. Agitation of the MNPs was performed at maximum speed to facilitate fast partitioning of the surfactant phase on the MNPs. Decreasing agitation speed from 2800 to 200 rpm had a negative effect on the extraction yield of the four compounds. Nevertheless, mixing time was a key factor determining the influence of agitation speed. In fact, decreasing vortex speed could be counterbalanced by an increase in mixing time; for the convenience of the analysis though, maximum mixing rate was applied. The desorption time necessary to back-extract the target analytes into the dichloromethane phase was subsequently tested (Fig. S3 – Supporting Information). Extraction was assisted by
3.3. Optimization of D--SPE conditions The first parameter investigated for its influence on the postCPE/D--SPE process was the amount of MNPs necessary to afford quantitative extraction. As shown in Fig. 6, increasing the amount of MNPs from 10 to 60 mg (2–12 mg/mL), the extraction efficiency was enhanced for the more hydrophobic EMC (log Kow = 5.80) and MBC (log Kow = 5.47) while the opposite trend was found for the less hydrophobic BZ3 (log Kow = 3.79) and BDM (log Kow = 4.51) [27]. Based on this pattern, 10 mg of the magnetic nano-sorbent was employed in the following experiments.
37
Fig. 7. Analyte signal response to extraction (mixing) time.
38
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
Table 1 Recoveries and relative standard deviation (n = 3) of the model organic compounds from spiked water samples. 50 ng/mL
100 ng/mL
River water
BZ-3 MBC BDM EMC
Lake water
River water
Lake water
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
Recovery (%)
RSD (%)
97.2 93.0 88.5 94.0
5.2 7.2 7.1 5.2
94.3 93.5 85.5 92.5
5.8 6.5 6.7 5.1
95.9 96.1 90.2 95.4
4.5 6.8 6.5 4.6
92.2 90.8 87.8 91.3
5.0 6.0 6.1 4.6
ultrasonic irradiation in order to accelerate analyte transfer [17,18]. The desorption/ultrasonication time was not found to have any influence on the analytical signal response above 30s. To ensure fast analysis time a desorption time of 30 s was therefore adopted. The desorption of the target analytes from the MNPs surface was investigated by varying elution solvent volume from 250 to 2000 L and for 1–5 runs. Increasing the elution volume was found to induce a decline in the analytical signals proportional to the theoretical preconcentration factor suggesting the lack of influence on the extraction performance. The use of lower solvent volume was not feasible because ultrasonication resulted in random scattering of the MNPs in the vial walls. Therefore, 250 L was decided as optimum. The peak intensities observed from 5 consecutive extractions revealed that two consecutive extractions are necessary to back-extract the analytes since approximately 10% of the analytes were detected in the second extraction step. Additional steps did not offer any improvement. The maximal sample volume that can be extracted with the proposed procedure was investigated by analyzing 10–100 mL of aqueous standard solutions considering that insufficient extraction yield occurs when it is decreases below 90%. This condition was satisfied up to 50 mL which is the maximum sample volume that can be extracted yielding theoretical preconcentration factors of approximately 100 (or higher with solvent evaporation). Extraction of larger sample volumes is feasible by appropriately scaling-up the experimental variables. This was evidenced by analyzing 100 mL sample volume with increased agitation time (2 min) and MNPs amount (30 mg). To ensure that the volume of CH2 Cl2 will not be a limiting factor, due to the increased amount of MNPs, it was adjusted to 1 mL (2× 500 L). Satisfactory extraction yields (88.7–93.1%) were observed suggesting that the method is flexible and can be tailored to cope with the requirements of the analysis. However, for the analytical demonstration purposes of this study, no further efforts were devoted to optimize the scaling-up factors of the experimental variables for the analysis larger sample volumes. The re-usability of the sorbent was finally assessed in a series of aqueous standard solutions (n = 5) extracted under the aforementioned procedure. After extraction, the MNPs were washed twice with ethanol and CH2 Cl2 under ultrasonic irradiation for 1 min. The solvent was removed by magnetic decantation and MNPs were dried in a ventilated oven at 50 ◦ C. The sorbents could be re-used for up to 4 times without impairing the extraction efficiency. It should be pointed out that MNPs re-usability strongly depended on the homogenization of the sorbent. Good homogenization was necessary to avoid agglomerates formed after drying. Finally, it should be emphasized that quantitative recovery of the sorbent was not feasible since a part strongly adhered to the vials after drying. Therefore, re-use after the fourth cycle offers insignificant gains with regard to sorbent consumption. On the basis of the above optimization study the total analysis time under the optimal extraction conditions for the two-step CPE-D--SPE process was less than 9 min which is a significant advantage over conventional CPE procedures where at least
30 min are required, depending also on the capacity of the centrifuge. 4. Analytical features and method application The analytical figures of merit of the proposed method were determined by using 10 mL aqueous standard solutions of the target analytes over a concentration range of 10–1000 ng/mL for BZ3, MBC and EMC and 50–1000 ng/mL for BDM. Method calibration was performed by plotting the mean peak area obtained from three measurements versus sample concentration. The regression coefficients varied between 0.990–0.997 and the detection limits, calculated as three times the signal-to-noise ratio were 1.43 ng/mL for BZ3, 2.4 ng/mL for MBC, 7.5 ng/mL for BDM and 1.94 ng/mL for EMC. Further improvement can be attained by resorting to the use of more sensitive analytical techniques, like LC–MS or GC–MS, by extracting larger sample volumes or by evaporating the extraction solvent to lower volume. However, both evaporation under N2 and at 50 ◦ C in a ventilated oven, were proven inefficient since an average loss of 30% in signal peak areas of BZ3, MBC and EMC, compared to direct analysis, was recorded; in addition, BDM peak disappeared. Mild evaporation of the elution solvent to dryness in a centrifugal concentrator under vacuum at room temperature was found to be the most appropriate technique, minimizing analytes’ loss and coping to a large extent with BDM evaporation. A series of experiments were then performed in order to assess the precision, repeatability and accuracy of the method. The repeatability of the method, expressed as intra-day relative standard deviation (RSD) of eight replicates was 4.5% for BZ3, 6.2% for MBC and 5.2% for EMC at 20 ng/mL and 7.5% and BDM at 50 ng/mL concentration level. The reproducibility of the measurements, expressed as inter-day RSD of five replicates was 8.7% for BZ3, 11.5% for MBC and 7.0% for EMC at 20 ng/mL and 14.9% for BDM at 50 ng/mL, which was deemed as satisfactory. Finally, the accuracy of the method was evaluated by spiking two genuine water samples at two different concentration levels (50 ng/mL and 100 ng/mL). The observed recoveries (Table 1) ranged between 85.5 and 97.2% suggesting that the method is tolerant of matrices generally representative of environmental water samples. 5. Conclusions In this study, a new two-step extraction technique combining CPE and D--SPE mediated by highly hydrophobic polysiloxanecoated core–shell Fe2 O3 @C magnetic nanoparticles is presented. On the basis of the results obtained the method affords significant simplification to conventional CPE procedures, without resource to appropriate or specialized apparatus, alleviating the need for specific sample handling treatment such as centrifugation or freezing of the surfactant-rich phase, significantly abridges the overall analysis time (at least 3-fold) and minimizes the importance of CPErelated parameters (i.e., surfactant concentration and incubation
D.L. Giokas et al. / J. Chromatogr. A 1251 (2012) 33–39
temperature) on the extraction. As an analytical demonstration, the parameters associated with method performance were optimized using four UV absorbing chemicals as model organic compounds and applied to spiked environmental water samples yielding good analytical features in terms of accuracy (85.5–97.2%), repeatability (4.5–7.5%) and reproducibility (7.0–14.9%). Considering the wide array of supramolecular solvents, offering a variety of functionalities and the high versatility of magnetic nanoparticles coatings, this study opens new venues in reinstating surfactant-mediated extractions. The use of ionic surfactant-mediated extractions that completely alleviate surfactant interference in chromatographic analysis, extraction of hydrophobic metal chelates and method automation are prosperous fields of research. Work along this line is in progress. Acknowledgements This research has been co-financed by the European Union (European Social Fund–ESF) and Greek national funds through the Operational Program “Education and Lifelong Learning” of the National Strategic Reference Framework (NSRF)–Research Funding Program: THALIS. Investing in knowledge society through the European Social Fund. The preparation of highly hydrophobic polysiloxane-coated core–shell Fe2 O3 @C magnetic nanoparticles was financially supported by the Natural Science Foundation of China (NSFC, Grant No. 50803013). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2012.06.054.
39
References [1] E.K. Paleologos, D.L. Giokas, M.I. Karayannis, Trends Anal. Chem. 24 (2005) 426. [2] L. Ramos, J. Chromatogr. A 1221 (2011) 84. [3] F. Sánchez-Rojas, C. Bosch-Ojeda, J.M. Cano-Pavón, Chromatographia 69 (2009) 79. [4] B. Hauser, M. Schellin, P. Popp, Anal. Chem. 76 (2004) 6029. [5] Spietelun, M. Pilarczyk, A. Kloskowski, J. Namie´snik, Chem. Soc. Rev. 39 (2010) 4524. [6] D. Han, K.H. Row, Microchim. Acta 176 (2012) 1. [7] M. Cruz-Vera, R. Lucena, S. Cardenas, M. Valcarcel, Anal. Methods 3 (2011) 1719. [8] N. Fattahi, S. Samadi, Y. Assadi, M.R.M. Hosseini, J. Chromatogr. A 1169 (2007) 63. [9] Q. Wua, C. Wang, Z. Liu, C. Wu, X. Zeng, J. Wen, Z. Wang, J. Chromatogr. A 1216 (2009) 5504. [10] Z-G. Shi, H.K. Lee, Anal. Chem. 82 (2010) 1540. [11] R.Z. Yang, J.H. Wang, M.L. Wang, R. Zhang, X.Y. Lu, W.H. Liu, J. Chromatogr. Sci. 49 (2011) 702. [12] H. Wang, H. Yan, J. Qiao, J. Sep. Sci. 35 (2012) 292. [13] A.B. Gómez, M.D. Sicilia, S. Rubio, Anal. Chim. Acta 677 (2010) 108. [14] R. Carabias-Martinez, E. Rodriguez-Gonzalo, B. Moreno-Cordero, J.L. PerezPavon, C. Garcia-Pinto, E. Fernandez Laespada, J. Chromatogr. A 902 (2000) 251. [15] Q. Zhu, F. Tao, Q.M. Pan, ACS Appl. Mater. Interfaces 2 (2010) 3141. [16] I.P. Roman, A. Chisvert, A. Canals, J. Chromatogr. A 1218 (2011) 2467. [17] D.L. Giokas, V.A. Sakkas, T.A. Albanis, D.A. Lampropoulou, J. Chromatogr. A 1077 (2005) 19. [18] T.I. Sikalos, E.K. Paleologos, Anal. Chem. 77 (2005) 2544. [19] R.P. Frankewlcht, W.L. Hinze, Anal. Chem. 66 (1994) 944. [20] D.L. Giokas, V.A. Sakkas, T.A. Albanis, J. Chromatogr. A 1026 (2004) 289. [21] L. Vidal, A. Chisvert, A. Canals, A. Salvador, Talanta 81 (2010) 549. [22] L.M. Grant, F. Tiberg, W.A. Ducker, J. Phys. Chem. B 102 (1998) 4288. [23] Y. Bai, D. Lin, F. Wu, Z. Wang, B. Xing, Chemosphere 79 (2010) 362. [24] Y. Geng, M.Y. Liu, J. Li, X.M. Shi, J.K. Kim, Comp. A: Appl. Sci. Manuf. 39 (2008) 1876. [25] M. Soria-Sanchez, A. Maroto-Valiente, A. Guerrero-Ruiz, D.M. Nevskaia, J. Colloid Interface Sci. 343 (2010) 194. [26] C.D. Stalikas, Trends Anal. Chem. 21 (2002) 343. [27] US EPA, Estimation Programs Interface SuiteTM for Microsoft® Windows, v 4.10, United States Environmental Protection Agency, Washington, DC, USA, 2011.